This makes it possible to access attributes and geometry from the parent
feature when in the filter of the "aggregate" expression function.
With this in place aggregates can be calculated per feature.
E.g. max "measurement" for each point_station per polygon_research_area.
Or a default attribute value when digitizing features:
aggregate(layer:='countries', aggregate:='max', expression:=\"code\",
filter:=intersects( $geometry, geometry(@parent) ) )
The order of the elements is irrelevant and duplicate elements are unwanted. It
is therefore a perfect candidate for a set instead of a list. This prevents
filtering for duplicates manually be replacing some filer codes with (more
performant) builtin methods of QSet.
Now all evaluate/prepare/etc methods must be called using QgsExpressionContexts
Also remove most remaining traces of special variables. This brings some
user facing changes, such that existing expressions may need to be
updated if they used these old special variables (eg $scale,
$feature). These changes are noted in doc/qgis3_user_changes.dox
so that we can include them in the release notes.
This change consolidates more unit handling and conversion into
QgsUnitTypes.
Additionally, UnknownUnit was renamed to UnknownDistanceUnit.
All methods which accepted QGis::UnitType parameters have been
updated to take QgsUnitTypes::DistanceUnit instead.
ALso remove the unit handling methods toLiteral, fromLiteral, tr,
fromTr, and fromUnitToUnitFactor from QGis. Their corresponding
counterparts in QgsUnitTypes should be used instead.
Move the QgsExpression::Interval class out to its own QgsInterval
class, extend with new methods and add tests
Add a typedef to keep API compatibility for 2.16
This commit sets the framework for allowing expression functions to
use named parameters. Ie, instead of:
clamp(1,2,3)
you can use:
clamp( min:=1, value:=2, max:=3)
This also allows arguments to be switched, eg:
clamp( value:=2, max:=3, min:=1)
Additionally, it allows for a more structured definition of function
parameters to handle optional arguments and default values for
parameters. These are currently being done using a hacky infinite
argument list.
I've utilised the postgres ':=' syntax for specifying named arguments
to avoid potential collisions which may arise with the equality test
if we re-used just the '=' operator alone.
Sponsored by North Road
calculations for $length, $area, $perimeter (refs #13209)
ie, the default is now to use planimeteric calculations unless
a geomCalculator has been explicitly set
copy/= operators or making them private
This revealed (and fixes) some issues, including a potential crash
using server access control (refs #13919), and a potential crash with
diagrams
Now all classes and members are either exposed to bindings or marked
as "not available in Python bindings" in the docs.
Drop test thresholds to 0. Now it should be much easier to determine
what missing members have been added which are causing test
failures.
This check tests that if a function has been declared deprecated
with either Q_DECL_DEPRECATED or has a @deprecated Doxygen note
then it MUST have both the Q_DECL_DEPRECATD and @deprecated note.
It's important that both are used, as Q_DECL_DEPRECATED allows
throwing a warning if that method is used in code, while the
@deprecated doxygen note gives an indication to devs/PyQGIS users
of why it's deprecated and what should be used instead.
Ideally we'd also test for SIP /Deprecated/ tags, but I can't
find any reliable way to do this.
Rationale:
- there was a lot of large objects passed by value, so potentially
there's a speed bump from this
- even for implicitly shared classes like QString/QList there's still
a (small) cost for copying the objects when there's no reason to
- it's the right thing to do!
This commit adds the ability for expressions to be evaluated against
specific contexts. It replaces the previous behaviour where
expressions were evaluated against a specific feature and could
utilise fragile global "special columns".
Now, expressions are instead evaluated using a context designed for
each individual expression. This is done via QgsExpressionContext
and QgsExpressionContextScope objects.
A QgsExpressionContextScope encapsulates the variables and functions
relating to a specific context. For instance, scopes can be created
for "global" variables (such as QGIS version, platform, and user-set
variables specified within the QGIS options dialog. Think things
like user name, work department, etc), or for "project" variables
(eg project path, title, filename, and user-set variables set
through the project properties dialog. Project version, reference
number, that kind of thing). Many more scopes are planned, including
map layer scopes (variables for layer name, id, user-set variables
through the layer properties dialog), composer scopes, etc...
QgsExpressionContextScopes are 'stacked' into a QgsExpressionContext
object. Scopes added later to a QgsExpressionContext will override
any variables or functions provided by earlier scopes, so for
instance a user could override their global 'author' variable set
within QGIS options with a different 'author' set via the project
properties dialog.
The intended use is that a QgsExpressionContext is created before
a batch set of QgsExpression evaluations. Scopes are then added to
the context based on what makes sense for that particular
expression. Eg, almost all contexts will consist of the global
scope and project scope, and then additional scopes as required.
So a composer label would be evaluated against a context
consisting of the global scope, project scope, composition scope
and finally composer item scope. The batch set of expression
evaluations would then be performed using this context, after which
the context is discarded. In other words, a context is designed
for use for one specific set of expression evaluations only.
